05 November 2019
The University of Leicester is renowned for its commitment to world-changing research.
If you want proof, just think back to February 2013, when it announced to the world’s press that its archaeological search in a Leicester car park led to the long-lost remains of King Richard III.
Founded in 1921, the university has three data centres all managed by Mark Penny and his team.
They are charged with providing backup for all home directories, corporate systems and research data (including HPC data) in a mixed environment encompassing Windows, VMWare and Lustre.
When an institution of this magnitude is entrusted with protecting the data of nearly 30,000 users, it requires a robust and dependable IT infrastructure.
It used to employ SAN-based hardware as backup targets for 10 media servers (each with 13x16 TB LUNs) with backup software provided by Commvault.
The index data was maintained by the individual media servers, so if one of them went down, it would disrupt the backup/restore process for that device, impacting all backup and restore operations.
What’s more, it would have resulted in zero access to backup data or restores, for as long as it took the IT team to acquire and install new hardware.
As you can imagine, this was a huge risk for University of Leicester and so a key objective was finding a solution that would allow the index data to be moved from the media servers to a robust, fault tolerant, shared storage environment, that same environment would also serve as the backup target.
Penny and his team evaluated SUSE Linux, Scality and Cloudian and these are his thoughts. “SUSE was massively more complex, with more hardware and we had significant concerns about the management difficulties this would create,” he explains. “Scality was also very complex, double the cost of Cloudian and required professional services to install. With Cloudian, I liked that I could try it in a VM and install it myself on my laptop in 15 minutes. It gave us confidence that we could easily manage the Cloudian system.”
After conducting a full proof concept using three Apollo 2U nodes loaned by HPE evaluation, Penny settled on Cloudian and so he and the IT crowd deployed Cloudian’s Hyperstore object storage system as the foundation of a revamped backup platform.
They deployed it on 12 HPE Apollo 2U servers with 3.4 PB raw storage.
For data protection, erasure coding was enabled in a 9+3 configuration, resulting in a 2.5 PB of usable capacity. In this configuration, up to three servers could fail simultaneously without impacting data availability.
Penny initially planned to deploy the system across the three data centres, but due to an unexpected surge in research data and the lack of additional funding to meet the unanticipated growth, he had to install the full system in just one site.
Now the university has not only eliminated the single point of failure, but also cut its storage space requirements by 50% and simplifies the backup process.
After fully implementing the Cloudian-based solution across its three data centres, the IT team expects to save in the region of 25% in data storage costs.
“We wouldn’t be able to continue backing up data, access backups or conduct restores until we replaced the hardware and spun up the system again, a process which could have taken weeks,” says Penny. ”Fortunately, we were never faced with this situation, but we knew we had to make a change.”